Transient Conduction:

Spatial Effects, Semi-Infinite Solids,

and Finite-Difference Methods

Chapter Five
Sections 5.4, 5.5, and 5.10
 Spatial Effects
• To use the lumped capacitance method for transient
conduction problems, we have to assume that the
temperature gradients within the medium are
negligible
• Now we consider problems in which we cannot
make this assumption (i.e. problems in which Bi >
0.1).
 Spatial Effects
• In their most general form, transient conduction
problems are described by the heat equation (see
Chapter 2)
• Remember – the heat equation is written differently
for different coordinate systems (Cartesian, radial,
spherical).
• The heat equations are partial differential equations.
The solutions provide the variation of temperature
with both time and spatial coordinates.
• However, in many problems (such as a plane wall),
only one spatial coordinate is needed to describe the
temperature distribution (1D)
 Spatial Effects
• For 1D Cartesian problems with no internal
generation and constant thermal conductivity, the
heat equation reduces to

• To solve Eq. 5.29 for the temperature distribution

T(x,t), we need an initial condition and two
boundary conditions.
• For a typical transient conduction problem, the
initial condition is given by an uniform initial
temperature distribution at time t = 0:
 Spatial Effects
• For the plane wall in Figure 5.4 (see earlier lecture
slides) the boundary conditions are at x = 0 and x =
L:

• The first boundary condition reflects the symmetry

requirement for the midplane of the wall (hence at the
midplane, the temperature gradient is 0).
• The second boundary condition describes the surface
condition experienced for time t > 0 (heat transfer by
conduction through the medium at x = L must equal heat
transfer convection out of the medium).
 Spatial Effects
• Using the heat equation and the equations for the
initial condition and the boundary conditions, we
can see that the temperatures in the wall depend on
a number of physical parameters:

• The problem we have described may be solved

analytically or numerically.
• In either case, there are advantages to
nondimensionalizing the governing equations.
 Spatial Effects
• To nondimensionalize the equations, we arrange the
relevant variables into suitable groups.
• First, consider the dependent variable T.
• If the temperature difference θ = T - T∞ is divided by
the maximum possible temperature difference θi =
Ti - T∞, a dimensionless form of the dependent
variable may be defined as

• Note: θ* must lie in the range 0 ≤ θ* ≤ 1.

 Spatial Effects
• Similarly, a dimensionless spatial coordinate may
be defined as

where L is the half-thickness of the plane wall

• Finally, a dimensionless time may be defined as

where t* is equivalent to the dimensionless Fourier

number (Fo = αt/L2)
 Spatial Effects
• Substituting Eqs. 5.34-5.36 into Eq. 5.29, the heat
equation becomes

and the initial and boundary conditions become:

where the Biot number is Bi = hL/k.

 Spatial Effects
• Recall that before we had a functional dependence
of

• Now that we have nondimensionalized the

equations, we can write the dimensionless form of
the functional dependence as

• Therefore, nondimensionalization has indeed

simplified the problem.
 Spatial Effects
• The Fourier number may be viewed as a
dimensionless time
• It also has an important physical interpretation
when used in problems for which heat transfer by
conduction through a solid is concurrent with
thermal energy storage by the solid.
• In these conditions, the Fourier number provides a
measure of the relative effectiveness with which a
solid conducts and stores thermal energy.
 The Plane Wall with Convection –
Exact Solution
• For a plane wall of thickness 2L, we can assume 1D
convection.
• If the wall is initially at a uniform temperature, T(x,
0) = Ti and is suddenly immersed in a fluid of T∞ ≠
Ti, the resulting temperatures may be obtained by
solving Equation 5.37 subject to the conditions of
Eqs. 5.38-5.40.
• An exact solution to this problem is of the form
 The Plane Wall with Convection –
Exact Solution

• Here, Cn is

and the discrete values of ζn (eigenvalues) are

positive roots of the equation
 The Plane Wall with Convection –
Approximate Solution
• It can be shown that for values of Fo > 0.2, the
infinite series solution (Eq. 5.42a) can be
approximated by the first term of the series.
• Hence, the temperature distribution becomes:

or

where represents the

midplane (x* = 0) temperature:
 The Plane Wall with Convection –
Approximate Solution

The coefficients C1 and ζ1 are evaluated from Eqs.

5.42b and 5.42c, respectively, and are given in Table
5.1 for a range of Biot numbers.
The Plane Wall with Convection –
Approximate Solution
The Plane Wall with Convection –
Approximate Solution
 The Plane Wall with Convection –
Approximate Solution

• The coefficients C1 and ζ1 are evaluated from Eqs.

5.42b and 5.42c, respectively and are given in Table
5.1 for a range of Biot numbers.

• Note that Table 5.1 also includes results for

transient 1D conduction for the infinite cylinder and
sphere. The exact solution and approximate
solutions for these systems can be found in a similar
manner to the method we used for the plane wall.
 The Semi-Infinite Solid
• Another simple geometry for which analytical
solutions may be obtained: the semi-infinite solid.
• In principle, this type of solid extends to infinity in
all but one direction
• This means it can be characterized by a single
identifiable surface (see Figure 5.7 on next slide)
• If we impose a sudden change of conditions at this
surface, transient, 1D conduction will occur
• The semi-infinite solid provides a useful
idealization for many practical problems, for
example heat transfer near earth’s surface or
transient response of a finite solid like a thick slab
The Semi-Infinite Solid
 The Semi-Infinite Solid
• The heat equation for transient conduction in a
semi-infinite solid is given by Eq. 5.29:

• The interior boundary condition is of the form

• Closed-form solutions for three important surface

conditions, instantaneously applied at t = 0, are
shown in Figure 5.7.
 The Semi-Infinite Solid
• We will consider case 1 in Figure 5.7: application of
a constant surface temperature.
• The solution may be obtained by recognizing the
existence of a similarity variable, η.
• We will use the similarity variable to transform the
heat equation from a partial differential equation
(with 2 independent variables, x and t) to an
ordinary differential equation expressed in terms of
the single similarity variable.
• This simplifies the equation and thus the problem!
 The Semi-Infinite Solid
• We’re going to cheat a bit. I will tell you that

and we will confirm this satisfies our requirements.

• First, we will transform the pertinent differential
operators:
 The Semi-Infinite Solid
• Substituting our results back into Eq. 5.29, the heat
equation becomes

• With x = 0 corresponding to η = 0, the surface

condition may be expressed as

• With x  ∞, as well as t  0, corresponding to η

 ∞, both the initial condition and interior
boundary condition correspond to a single
requirement that
 The Semi-Infinite Solid
• Since the transformed heat equation and the
initial/boundary conditions are independent of x and
t, we have now confirmed that

is a similarity variable. Its existence implies that,

irrespective of the values of x and t, the temperature
may be represented as a unique function of η.
 The Semi-Infinite Solid
• The specific form of the temperature dependence,
T(η), may be obtained by separating variables in
Eq. 5.57:

• Integrating, it follows that

or
 The Semi-Infinite Solid
• Integrating a second time, we obtain

where u is a dummy variable. Applying the

boundary condition at η = 0, it follows that C2 = Ts and
 The Semi-Infinite Solid
• From the second boundary condition, we obtain

or, evaluating the definite integral,

• Hence, the temperature distribution may be

expressed as
 The Semi-Infinite Solid

• Here, the Gaussian error function, erf η, is a

standard mathematical function that is tabulated in
Appendix B.
• Erf(η) asymptotically approaches unity as η
becomes infinite.
• Thus, at any nonzero time, temperatures
everywhere are predicted to have changed from Ti
(and to have become closer to Ts).
 The Semi-Infinite Solid
• Now that we have a temperature distribution, the
surface heat flux may be obtained by applying
Fourier’s law at x = 0, in which case

• Analytical solutions may also be obtained for the Case 2 and

Case 3 surface conditions in Figure 5.7.
• Solutions for all three surface condition cases in Figure 5.7
are summarized in the following slides
 The Semi-Infinite Solid
• Case 1 - Constant Surface Temperature: T(0,t) =
Ts
 The Semi-Infinite Solid
• Case 2 - Constant Surface Heat Flux:
 The Semi-Infinite Solid
• Case 3 – Surface Convection:

Here, the complementary error function, erfc w, is

defined as erfc = 1 – erf w.
 Finite-Difference Methods
• Up to this point, we’ve studied only analytical
solutions to transient problems
• Analytical solutions to transient problems are
restricted to simple geometries and boundary
conditions, (like the 1D cases we’ve been studying)
• However, in many cases geometry and/or boundary
conditions preclude the use of analytical techniques
• In these cases, we can use finite-difference (or
finite-element) methods
• Here, we will consider both explicit and implicit
forms of finite-difference solutions to transient
problems
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Let us again consider the 2D system of Figure 4.4
(see next slide).
• Under transient conditions with constant properties
and no internal generation, the appropriate form of
the heat equation is
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• To obtain the finite-difference form of this equation,
we may use the central-difference approximations
to the spatial derivatives prescribed by Eqs. 4.27
and 4.28.
• Once again, the m and n subscripts may be used to
designate the x and y locations of discrete nodes.
• However, in addition to being discretized in space,
this problem must also be discretized in time.
• The integer p is introduced for this purpose, where
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Thus, the finite-difference approximation to the
time derivative in Eq. 5.75 is expressed as

• The superscript p denotes the time dependence of T

• The time derivative is expressed in terms of the
difference in temperatures associated with the new
(p + 1) and previous (p) times.
• Hence, calculations must be performed at
successive times separated by the interval Δt
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Note: just as finite-difference solutions restrict
temperature determination to discrete points in
space, it also restricts it to discrete points in time
• Back to derivation: if Eq. 5.77 is substituted into
Eq. 5.75, the nature of the solution will depend on
the specific time at which temperatures are
evaluated in the finite-difference approximations to
the spatial derivatives.
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• In the explicit method of solution, these
temperatures are evaluated at the previous (p) times.
• Hence, Eq. 5.77 is considered to be a forward-
difference approximation to the time derivative.
• Evaluating terms on the right-hand side of Eqs. 4.27
and 4.28 at p and substituting into Eq. 5.75, the
explicit form of the finite-difference equation of the
interior node m, n is
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Solving for the nodal temperature at the new (p + 1)
time and assuming that Δx = Δy, it follows that

where Fo is the Fourier number

 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• This approach can easily be extended to 1D or 3D
systems. If the system is 1D in x, the explicit form
of the finite-difference equation for an interior node
m reduces to

• Eqs. 5.79 and 5.81 are explicit because unknown

nodal temperatures for the new time are determined
exclusively by known nodal temperatures at the
previous time.
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Since the temperature of each interior node is known at
t = 0 (p = 0) from prescribed initial conditions, the
calculations begin at t = Δt (p = 1), where Eq. 5.79 or
5.81 is applied to each interior node to determine its
temperature.
• With temperatures now known for t = Δt, the
appropriate finite-difference equation is then applied at
each node to determine its temperature at t = 2Δt (p =
2).
• In this way, the transient temperature distribution is
obtained by marching out in time using intervals of Δt.
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Note: the accuracy of the finite-difference solution
may be improved by decreasing the values of Δx
and Δt.
• However, decreasing either or both of these
increases computation time.
• There are also stability requirements that must be
considered when choosing Δt
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• The explicit method is not unconditionally stable –
the solution may not always continuously approach
final (steady-state) values with increasing time.
• Instead, with the explicit (but not implicit) method,
the solution may be characterized by numerically
induced oscillations, which are physically
impossible.
• These oscillations may become unstable, causing
the solution to diverge from the actual steady-state
conditions.
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• To prevent such erroneous results, the prescribed
value of Δt must be maintained below a certain
limit, which depends on Δx and other system
parameters
• This dependence is called a stability criterion
• For the problems of interest here, the criterion is
determined by requiring that the coefficient
associated with the node of interest at the previous
time is greater than or equal to zero.
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• In general, this is done by collecting all terms
involving to obtain the form of the coefficient.
• The result is then used to obtain a limiting relation
involving Fo, from which the maximum value of Δt
may be determined.
• For example: with Eqs. 5.79 and 5.81 already
expressed in the desired form, it follows that for a
2D note, the stability criterion is
• or
• And for a 1D interior node it is
• or
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Equations 5.79 and 5.81 may also be derived by
applying the energy balance method to a control
volume about the interior node.
• Accounting for changes in thermal energy storage, a
general form of the energy balance equation may be
expressed as

where it is assumed that all heat flow is into the node.

 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Example: consider the surface node of the 1D system
in Figure 5.12:
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• To more accurately determine thermal conditions
near the surface, this node has been assigned a
thickness that is one-half that of the interior nodes.
• Assuming convection transfer and no generation, it
follows from Eq. 5.84 that:

or, solving for the surface temperature at t + Δt:

 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method

• Recognizing that

and rearranging, it follows that

 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
 

• Recalling the procedure for determining the stability

criterion, we require that the coefficient for be
greater than or equal to 0. Hence,

or
 Finite-Difference Methods
Discretization of the Heat Equation: the Explicit Method
• Note that the complete finite-difference solution
requires the use of Eq. 5.81 for the interior notes and
Eq. 5.85 for the surface node
• Thus, Eq. 5.87 must be contrasted with Eq. 5.82 to
determine which stability requirement is more
stringent.
• Since Bi ≥ 0, the limiting value of Fo for Eq. 5.87 is
less than that for Eq. 5.82.
• Eq. 5.87 should therefore be used to select the
maximum allowable value of Fo and hence Δt to be
used in the calculations.
 Finite-Difference Methods
Discretization of the Heat Equation: the Implicit Method
• In the explicit finite-difference scheme, the temperature
of any node at t + Δt may be calculated from knowledge
of temperatures at the same and neighboring nodes for
the preceding time t.
• Hence, determination of a nodal temperature at some
time is independent of temperatures at other notes for the
same time.
• However, as mentioned in the last slide, this method has
limitations for the time interval Δt if a stable solution is to
be reached.
• This often results in small Δt and large computation
time
 Finite-Difference Methods
Discretization of the Heat Equation: the Implicit Method
• A reduction in computation time may often be
realized by employing an implicit, rather than
explicit, finite-difference scheme.
• Implicit finite-difference schemes do not require a
stability criterion (they are unconditionally stable)
and thus allow larger Δt, reducing computation time.
• In the implicit scheme, all other temperatures are
evaluated at the new (p + 1) time rather than the
previous (p) time.
• Eq. 5.77 is then considered to provide a backward-
difference approximation to the time derivative.
 Finite-Difference Methods
Discretization of the Heat Equation: the Implicit Method
• In contrast to Eq. 5.78, the implicit form of the
finite-difference equation for the interior node of a
2D system is then
 Finite-Difference Methods
Discretization of the Heat Equation: the Implicit Method
• Rearranging and assuming Δx = Δy, it follows that

• From Eq. 5.95, it is evident that the new temperature

of the m, n node depends on the new temperatures of
its adjoining nodes, which are, in general, unknown.
• Hence, to determine the unknown nodal
temperatures at t + Δt, the corresponding nodal
equations must be solved simultaneously.
 Finite-Difference Methods
Discretization of the Heat Equation: the Implicit Method
• Unlike the explicit method, the implicit formulation
is unconditionally stable.
• That means that there are no restrictions on Δx or Δt.
• Since larger values of Δt may be used, computation
times may often be reduced by using the implicit
method, with little loss of accuracy.
• Still, to maximize accuracy, Δt should be
sufficiently small to ensure that the results are
independent of further reductions in its value
(similar to a grid conversion test for Δx, we need to
perform a Δt conversion test)